基于最优内插预测的科学数据压缩方法

Scientific Data Compression Method Based on Optimized Interpolate Predition

在海量科学数据存储和传输压力愈来愈大的背景下,我们针对结构网格离散的科学计算数据,研究了基于二维9点、三维27点最优内插预测的科学数据压缩方法。数值实验表明该方法大大优于现有的压缩算法,可以较好地解决科学计算数据的压缩存储问题。

Application in scientific computing often operate with large volume of output and input data, the space required for data storage and data transfer over communication networks are bottleneck in scientific computing. In the paper,we present a compression method based on optimized prediction operator which applied to structured scientific data.the proposed approach may be used both for lossy and lossless compression and is well suited for scalar fields. Numerical results indicate that our method successfully outperforms universal compression algorithm greatly.